Six IC projects gain funding from Microsoft

© 2016 EPFL

© 2016 EPFL

The Swiss Joint Research Centre, a collaborative research engagement between Microsoft Research, EPFL and ETHZ, has awarded funding to six IC and four ETHZ projects, on the basis of intellectual merit, potential impact as well as relevance, for a period of up to two years beginning in 2017.

The five collaborative project proposals, which successfully obtained the two-year funding, are the following:

“Near-Memory System Services”, has the goal of satisfying the performance requirements of data center services at a fraction of a modern infrastructure’s power. The proposal was created by Babak Falsafi (IC) and Stavros Volos (MSR Redmond).

“Coltrain: Co-located Deep Learning Training and Inference”, a project proposal developed by Babak Falsafi, Martin Jaggi (IC) and Eric Chung (MSR Redmond), which aims to address the fundamental challenges in Deep Neural Network (DNN) inference/training co-location on servers with integrated FPGAs.

“From Companion Drones to Personal Trainers”, developed by Pascal Fua, Mathieu Salzmann (IC), Debadeepta Dey, Ashish Kapoor, and Sudipta Sinha (MSR Redmond). The goal of the project will be to replace the GoPro-style action cameras by drones using an on-board camera as a sensor and to film the person during an exercise routine adding useful advice to the sequences.

“Revisiting Transactional Computing on Modern Hardware”, the project - created by Rachid Guerraoui, Georgios Chatzopoulos (IC) and Aleksandar Dragojevic (MSR Cambridge), will focus on the design and building of support for high-level transactions on top of modern hardware platforms, using the Structured Query Language (SQL).

“Fast and Accurate Algorithms for Clustering”, developed by Michael Kapralov and Ola Svensson (IC), Yuval Peres and Konstantin Makarychev (MSR Redmond), has the goal of designing faster and more accurate algorithms for fundamental clustering problems. Moreover, the project aims to design algorithms that achieve optimal trade-offs between approximation quality, runtime and space-communication complexity, making them well-suited for modern data models.

“Towards Resource-Efficient Data Centers”, a project which aims to improve the utilization of server resources in data centers led by Florin Dinu (IC), Christos Gkantsidis and Sergey Legtchenko (MSR Cambridge), has gained an additional year of support.

IC School congratulates all the successful Project Leaders.